General Concepts & Functionality
Q: How is Rebel’s memory different from standard chat history?
Unlike simple chat history, Rebel uses a structured knowledge architecture designed to help the AI maintain context about your work and preferences without overloading every conversation. It acts as a routing and retrieval system, knowing where to look for information rather than just storing a long transcript of everything you have ever said.
Q: Does Rebel load everything I’ve ever told it into every conversation?
No. Rebel uses a two-tier memory system to manage context:
• Tier 1 (Auto-loaded): Essential facts about your identity, preferences, and current focus are stored in README.md files. This information is loaded automatically into every conversation,.
• Tier 2 (On-demand): Detailed context, such as project histories or client details, is stored in memory/topics/. Rebel searches for and loads these only when you ask about them,.
Q: How does Rebel decide what to auto-load versus what to store for later?
It is based on utility. Information that is needed in more than 50% of your interactions (like your role or timezone) goes into the auto-loaded README.md,. Detailed information needed less frequently (less than 50% of the time) is stored in topic files,.
Privacy & Sharing
Q: Can my team see everything I save to Rebel’s memory?
No. Rebel organizes memory into two distinct privacy zones:
• Chief-of-Staff/: This is your private space. It is never shared with anyone and is meant for personal info, sensitive matters, and private preferences,.
• work/Company/: This is a shared workspace visible to your team. Use this for client context, project docs, and team workflows,.
Q: How do I know where to save a specific piece of information?
You can use a simple decision flow:
1. Is the information sensitive, personal (e.g., salary, health), or a private reflection? If yes, it goes to Chief-of-Staff/.
2. Should your team have access to this (e.g., project status, client history)? If yes, it goes to work/Company/. If no, keep it in Chief-of-Staff/.
Q: Can I keep private notes inside a shared work file?
Yes. Rebel supports Sensitivity Markers. Within a single file, you can mark specific sections as private using ## PERSONAL (for things like private concerns) and others as ## SHAREABLE (for team visibility). This allows you to maintain personal context alongside team-visible information in the same document.
Data Storage & Integrity
Q: Does Rebel copy the contents of my emails and files into its memory?
Generally, no. Rebel follows a “Pointers, Not Copies” mental model. Instead of duplicating content, memory files act as a library catalogue containing query strings or links that point to the Single Source of Truth (SSOT), such as a specific Gmail search or a Notion URL. This keeps memory lightweight and prevents information from going stale.
Q: How does Rebel handle sensitive financial or health information?
You can set explicit rules in your private README.md, such as “Compensation information is PERSONAL.” Rebel follows these rules to ensure sensitive data is never saved to shared memory spaces. If Rebel is unsure, it is designed to ask you for permission before saving potentially sensitive information to shared spaces.
Analogy for Better Understanding
To understand how Rebel manages memory, think of it like a computer’s hardware:
• The Auto-loaded Tier (README.md) is like RAM. It is fast and always active, but it has limited space, so you only keep the most essential, frequently used data there (like your operating system or open apps).
• The On-demand Tier (memory/topics/) is like a Hard Disk. It has massive storage capacity for all your files and projects, but the computer only retrieves specific files when you actually open them.